Skip to contentSkip to conversation

Heavy disk IO stalls ftp/http downloads

DISCUSSION Samium Gromoff 2 participants 2 messages Nov 23, 2001

lore
Share & tools
lore
#0 lore raw

does hdparm -u1 fixes the problem?
-u1 == unmasq irq during its handling
the problem is that serial port need its interrupt to be served with minimal
latency, and the drives makes that impossible, unless irqs are unmasked...

looks like ( = - disc io irq handling periods, . - the ones of serial port):

both ones wants something like:

disc: ========= ============= ================== ============= =========
serial: . . . . . . . . . . . . . . . . . . . . . . .

in reality: ==============.=============. ============= ===========

etc etc...

cheers, Samium Gromoff

#1 lore raw
Quoted 5 lines · Samium Gromoff

does hdparm -u1 fixes the problem?
-u1 == unmasq irq during its handling
the problem is that serial port need its interrupt to be served with
minimal latency, and the drives makes that impossible, unless irqs are unmasked...

Hello Samium,

Thanks a lot for your response.

Yes, I have -u1 support enabled for both the hard drives (in fact it is /sbin/hdparm -m16 -c3 -u1 -X66 -d1 -W1 /dev/hda and /dev/hdc). But, unfortunately it does not fix the problem.

Here is the 'hdparm' output from the hard drives:
/dev/hda:

 multcount    = 16 (on)
 I/O support  =  3 (32-bit w/sync)
 unmaskirq    =  1 (on)
 using_dma    =  1 (on)
 keepsettings =  0 (off)
 nowerr       =  0 (off)
 readonly     =  0 (off)
 readahead    =  8 (on)

geometry = 1247/255/63, sectors = 20044080, start = 0

/dev/hdc:

 multcount    = 16 (on)
 I/O support  =  3 (32-bit w/sync)
 unmaskirq    =  1 (on)
 using_dma    =  1 (on)
 keepsettings =  0 (off)
 nowerr       =  0 (off)
 readonly     =  0 (off)
 readahead    =  8 (on)

geometry = 1240/255/63, sectors = 19931184, start = 0

Signature · 2 lines
Hari.
[email protected]